Non-vertebral fracture risk reduction with oral bisphosphonates: challenges with interpreting clinical trial data

To license a therapy for the treatment of postmenopausal osteoporosis pharmacological agents must show ability to reduce the incidence of morphometric vertebral fractures versus placebo over a 3-year study period. In Europe, recent registration guidelines require evidence of reduction of vertebral and non-vertebral fracture incidence over a minimum of 2 years compared with placebo. There is much interest in the prevention of non-vertebral fractures. While morphometric vertebral fractures are assessed and statistically powered as the registration primary endpoint in clinical trials, non-vertebral fractures are often measured as secondary endpoints or captured as adverse events, which have selection biases in data capturing.

To describe factors that influence fracture risk and the rate of osteoporotic non-vertebral fractures observed in randomized controlled studies of the oral nitrogen-containing bisphosphonates licensed for the treatment of postmenopausal osteoporosis (alendronate, risedronate and ibandronate).

A literature search was conducted using PubMed and ISI Web of Knowledge and using keywords representing drug names and trial types. Results were screened using selection criteria based on trial type and vertebral fracture endpoint of trials published from 1990 to 2007.

Without comparative head-to-head antifracture studies, current evidence does not support a clear differentiation in fracture reduction among the different bisphosphonates. The rate of fracture in a clinical study is dependent on different factors (e.g., skeletal fragility), which may vary from study to study. Even in trials assessing non-vertebral fractures as a primary endpoint, differences in study design, randomized population and varying definitions of what constitutes a non-vertebral fracture can influence outcomes. In addition, falls and fall-related risk factors have never been controlled for in or between individual studies. Although etidronate, administered with an extended between-dose interval, has demonstrated a significant reduction in fracture risk, this was in a subgroup population, with the addition of phosphate or when only data from weeks 61-150 of the study were included in the analysis. None of the remaining currently registered non-daily oral bisphosphonates have prospective fracture data, and have, therefore, been registered on the basis of non-inferiority (surrogate marker bone mineral density and bone turnover marker) endpoints. However, a lack of evidence, if the appropriately designed studies have not been completed, does not necessarily indicate a lack of efficacy. Such a conclusion can only be drawn if a suitable study has been completed that definitively shows a lack of effect on non-vertebral fractures.